Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Zhang Boyang <zhangboyang.id@gmail.com>
To:     Andrey Grodzovsky <andrey.grodzovsky@amd.com>,
        =?UTF-8?q?Christian=20K=C3=B6nig?= <christian.koenig@amd.com>,
        linux-kernel@vger.kernel.org, amd-gfx@lists.freedesktop.org
Cc:     "David C . Rankin" <drankinatty@suddenlinkmail.com>,
        Steven J Abner <pheonix.sja@att.net>
Subject: [RFC PATCH 0/1] drm/amdgpu: Fix NULL-deref in amdgpu_device_fini_sw()
Date:   Sat,  1 Oct 2022 05:41:09 +0800
Message-Id: <20220930214110.1074226-1-zhangboyang.id@gmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi,

There are several reports of "Fatal error during GPU init" will cause
NULL-deref in amdgpu_device_fini_sw(). Although the NULL-deref is result
instead of reason, this NULL-deref will confuse user.

https://lore.kernel.org/lkml/a8bce489-8ccc-aa95-3de6-f854e03ad557@suddenlinkmail.com/
https://lore.kernel.org/lkml/AT9WHR.3Z1T3VI9A2AQ3@att.net/

This is probably because "adev" is not fully initialized when
amdgpu_device_init() failed. Thus subsequent amdgpu_device_fini_sw()
will try to release "adev->reset_domain" and cause NULL-deref.

This patch fixes this problem by guarding the code with an "if".
However, I'm new to this module and I didn't fully understand the code,
so please review my code carefully.

Best Regards,
Zhang Boyang